EBSD Study of Damage Mechanisms in a High-StrengthFerrite-Martensite Dual-Phase Steel

N. Saeidi, F. Ashrafizadeh, B. Niroumand, and F. Barlat

(Submitted January 31, 2014; in revised form September 3, 2014; published online October 15, 2014)

Electron backscattered diffraction (EBSD) analyses were performed on a fine-grained dual-phase (DP)sheet steel subjected to uniform tensile deformation and the preferred void nucleation sites as well as themicro-mechanisms of void formation were examined. EBSD study of grain average misorientation, grainorientation spread and kernel average misorientation of the deformed microstructure revealed that voidsnucleation initially happened at ferrite-martensite interfaces neighboring rather large ferrite grains. This isbelieved to be mainly due to the higher shear deformation ability of the larger ferrite grains, the highernumber of dislocation pile-ups at the martensite particles and the less uniform strain distribution within thelarger ferrite grains compared to the smaller ones. The results demonstrated the impact of increasinguniform strain distribution within the DP microstructure on lowering the void nucleation probability.

Keywords dual-phase steel, EBSD analysis, void nucleationmicro-mechanisms

1. Introduction

Ductile fracture of metals and alloys generally occursthrough the following sequence: void nucleation, growth, andcoalescence (Ref 1). Voids can be initiated either by thecracking of particles or the separation of particle/matrixinterfaces (Ref 2–4). It has been shown (Ref 3) that in ferrite-pearlite microstructures, the formation of microcracks atcementite platelets in pearlite and the initiation of voids alongwell-developed dislocation cell walls in ferrite grains are themain void nucleation mechanisms. Since large variations inconstituent particles morphology normally exist simultaneouslyin the engineering alloys, the nucleation behavior of voids israther complicated and accordingly, each specific alloy systemshows its characteristic behavior. Studying different martensitemorphologies and their distribution in ferrite-martensite DPsteels, Cingara et al. (Ref 5) showed that the uniformdistribution of martensite particles within ferrite matrix dis-couraged void growth; however, it encouraged continuous voidnucleation up to the final fracture. Also, they found that at strainvalues high enough to approach the fracture strain, the mainpopulation of voids was formed by the decohesion of ferrite-martensite interfaces (Ref 6).

Void initiation between closely situated martensite particlesalong ferrite grain boundaries, in dual-phase steels, was also

observed by Kadkhodapour et al. (Ref 7). They ascribed thisphenomenon to the strength mismatch between ferrite andmartensite grains that made local stresses perpendicular to theloading direction. So, this led to the decohesion of ferrite-ferritegrain boundaries in the neighborhood of martensite grains andthe creation of void starts from the early stages of deformation.

Moreover, it was shown that a coarse and interconnectedmartensite distributed along ferrite grain boundaries was proneto easily crack at lower strains followed by the decohesion ofthe ferrite-martensite interface at higher strains (Ref 6). In thecase of a fine structure, martensite cracking was less frequentand void formation by ferrite-martensite interface separationwas the dominant mechanism (Ref 8). Recently, the strainpartitioning among the martensite and ferrite of DP steels hasbeen directly measured and proved by in situ tensile testscoupled with digital image correlation or micro-grid techniques(Ref 9); it was found that the strain partitioning between themartensite and ferrite was a key factor controlling the voidformation behavior in DP steels, depending mostly on thevolume fraction and hardness of martensite particles.

Although the void nucleation mechanism by ferrite-mar-tensite interface decohesion is well established, importantparameters encouraging this kind of void initiation at specificregions have received less attention. In most cases, it has beensimply stated (Ref 10, 11) that void nucleation happens atmartensite particles located at the ferrite grain boundaries, butthe details of the process have not been clarified. In the presentstudy, by using detailed SEM and EBSD analysis of tensiletested DP780 sheet steel specimens, void creation in a ferrite-martensite DP microstructure was examined and its correlationwith the ferrite grain size was identified.

2. Materials and Methods

DP780 sheet steel used in this research was provided byPOSCO Company, South Korea. Tensile specimens weremachined according to ASTM A370 standard (Ref 12) in the

N. Saeidi, F. Ashrafizadeh, and B. Niroumand, Department ofMaterials Engineering, Isfahan University of Technology, Isfahan84156-83111, Iran; and F. Barlat, Materials Mechanics Laboratory(MML), Graduate Institute of Ferrous Technology (GIFT), PohangUniversity of Science and Technology (POSTECH), San 31 Hyoja-dong, Nam-gu, Pohang, Gyeongbuk 790-784, Republic of Korea.Contact e-mail: navidsae@gmail.com.

JMEPEG (2015) 24:53–58 �ASM InternationalDOI: 10.1007/s11665-014-1257-4 1059-9495/$19.00

Journal of Materials Engineering and Performance Volume 24(1) January 2015—53

rolling direction (Fig. 1), using electrical discharge machining(EDM) method. The sheet thickness was about 1 mm. The gagelength was 50 mm and the tensile tests were performed at aconstant cross-head speed of 1 mm/min with a servo-hydraulicMTS machine. The deformed microstructures of the specimenswere examined by a scanning electron microscope (SEM)equipped with an EBSD detector. The specimens were thensectioned through thickness along the mid-width in thelongitudinal direction using Struers cutting machine. In orderto measure local strains during deformation, these sectionedspecimens were mounted, ground, and polished to 4000 gritfinish and polished with 1-micron diamond suspension. Thespecimens were then polished using a colloidal silica suspen-sion which slightly etched the surface of the material. Someother specimens were etched in 2% Nital solution. Character-istics of void features across the length of the specimen werethen studied by SEM and EBSD. EBSD scans were performedon a TSL system mounted on a Zeiss ULTRA 55 microscope.Image quality (IQ), grain average misorientation (GAM), grainmisorientation spread (GOS), and Kernel average misorienta-tion (KAM) maps were mainly investigated by EBSD analysis.IQ value represents the sharpness of Kikuchi patterns at anygiven point. GAM value is an indication of the averageneighbor-to-neighbor misorientation within the grains in theactive partition. GOS value of a point indicates the orientationspread of the grain to which the point belongs. KAM value foreach pixel is defined as the average misorientation that a pixelhas with its neighbors (6 in our case), while boundaries withorientation difference over 5� are recognized as grain bound-aries. KAM value is an appropriate quantity used to evaluatethe strain or the stored energy at a given point (Ref 13, 14).

3. Results and Discussion

The initial microstructure of the studied steel is presented inFig. 1. As shown, the distribution of martensite particles wasnearly uniform within the microstructure of DP780 steel and nonoticeable banding could be observed. The yield strength ofthis steel was obtained to be equal to 550 MPa. Also, theultimate tensile stress and failure strain were found to be893 MPa and 0.24, respectively (Fig. 2).

Detailed SEM observations showed that most initial voidswere nucleated between closely spaced martensite particles(Fig. 3) located between rather two large ferrite grains. In otherwords, the area with the lower ratio or qm/qf, in which qm is thelocal density of martensite particles and qf is the local densityof voids, is more susceptible to void creation. A straightforwardanalysis for the confirmation of this statement is to use theQuadrat method (Ref 15). A schematic representation of thismethod is shown in Fig. 4. For this analysis, multiple SEM

micrographs, up to 10 micrographs and from the same strainvalue, were divided into a grid of square cells. Then, thenumber of martensite particles as well as voids in each cell wascounted and divided by each cell area. So, the density ofmartensite particles and voids in different regions weremeasured. Consequently, the relative martensite occupation(qr) for each cell was defined as qr = qm/qf. Cumulativeanalysis of relative martensite occupation data was conductedas presented in Fig. 5. It was evident that the frequency ofoccurrence in low values of qr, i.e., the lower number ratio ofmartensite particles to the voids, was much higher than theother values. In other words, occurrence of void creation inareas with low density of martensite particles was moreprobable than the areas with high density of martensiteparticles.

In order to further explore this issue, EBSD analysis wasperformed on the deformed microstructure. Figure 6 showsimage quality (IQ) map of the microstructure. Dark grains inthe IQ map incorporated the martensite phase in which, due tothe highly distorted lattice, Kikuchi patterns appeared faint oreven invisible. Because of the absence of Kikuchi patterns,voids appeared very dark in this map; arrows in the IQ mapindicated the voids. By the further processing of EBSD data,grain orientation spread (GOS) map, grain average misorien-tation (GAM) map, and Kernel average misorientation (KAM)map were also obtained. GAM map of the microstructure ispresented in Fig. 7. It can be seen that this parameter was not sodifferent within the constituent grains; it is also evident thatGAM was nearly the same for grains surrounding the voids,such as grains shown as A and B (Fig. 7). According to KAMmap (Fig. 8), the local misorientation at grain boundaries washigher than that within the grains. During uniaxial straining, thetendency toward localized inhomogeneous deformation due to

Fig. 1 Geometry of the tensile testing specimen

Fig. 2 Initial microstructure of DP780 steel

54—Volume 24(1) January 2015 Journal of Materials Engineering and Performance

the different mechanical behavior of ferrite and martensite, aswell as some restriction in shear deformation of ferrite grains bymartensite particles, led to the creation of a high density ofgeometrically necessary dislocations, especially near the inter-face of ferrite and martensite (Ref 16, 17). So, in fine-grained

microstructures incorporating a high density of grain bound-aries or interfaces, geometrically necessary dislocations shouldbe higher within most parts of the grain. As a result, straindistribution would be more uniform within the grains in a fine-grained microstructure than that in a coarse grained one. It canbe said that the lesser the grain boundary to grain area ratio, thelower is the area fraction of the grain incorporating high KAMvalues; that is, area fraction of high KAM to low KAM valuesin large grains could be lower and consequently, GAM value ofthe grain would be lower in large ferrite grains than that of finegrains, as can be seen in Fig. 7. It is worth noting that straindistribution mentioned referred to the local strain within themicrostructure. Owing to different strength properties of ferriteand martensite phases in a dual-phase steel, they could not bedeformed uniformly during tensile test; therefore, immediatelyafter yielding, the ferrite phase was deformed while martensitedeformation would occur at higher strains and at a much lowerrate. So, the strain distributions between ferrite and martensitegrains would be inhomogeneous (Ref 7). It can be concludedthat the higher density of geometrically necessary dislocationswas created near the interfaces of ferrite and martensiteparticles (Ref 18).

In smaller ferrite grains (for example, grain C in Fig. 7),compared to the larger grains, a higher fraction of boundary wascovered by martensite particles. Therefore, the restricting effectof hard martensite particles on the deformation of small ferritegrains was higher than that of the larger grains. On the otherhand, as observed in the larger grains (for example, A and B inFig. 7), shear deformation of these grains was less restricted bythe surrounding martensite particles. Thus, the larger the ferritegrain size, the higher the shear deformation that could besustained; therefore, in large grains, shear deformation bands canbe observed in slightly etched specimens (Fig. 9).

Fig. 3 Void nucleation between martensite particles located at theferrite grain boundaries, by interface decohesion

Fig. 4 Schematic presentation of Quadrat method analysis, themicrograph were divided by 9 cells, the voids as well as martensiteparticles was indicated with arrows

Fig. 5 Frequency distribution of relative martensite occupation (qr)

Fig. 6 IQ map of deformed microstructure

Journal of Materials Engineering and Performance Volume 24(1) January 2015—55

GOS map (Fig. 10) shows that in grains having nearly thesame GAM values (Fig. 7), there existed different orientationspreads, especially in grains A and B which surrounded theindicated voids in the map. The large grains of A and B had the

same GAM value, but they showed different orientationspreads, GOS values. By considering GOS map (Fig. 10), itcan be seen that there was more orientation spread in the largergrains compared to the smaller grains. GOS value could be an

Fig. 7 GAM map of deformed microstructure

Fig. 8 KAM map of deformed microstructure

56—Volume 24(1) January 2015 Journal of Materials Engineering and Performance

indication of strain distribution within the grain such that itcould be expected that the lower orientation spread in a grainwould result in better strain distribution within the grain. Betterstrain distribution within a small ferrite grain than a large ferritegrain could be justified by considering the constraint effect ofthe neighboring martensite particles on ferrite grains whichimposed back stresses within the ferrite grain. It was shown thatdecreasing the grain size increased the back stress imposed bythe stored dislocations at the dislocation emission points (Ref19). Moreover, grain boundary martensite particles couldaccentuate the mentioned back stress (Ref 20). Therefore, it

could be said that in the DP steel studied, the smaller the grainsize, the higher the back stress created within the grain, per unitvolume; therefore, the stress needed to generate successivedislocations from a given source was increased. As a conse-quence, in small grains, to continue the deformation, moredislocation sources should be activated. It was expected that thecreated dislocations from different sources would have auniform distribution encouraging more uniform strain distribu-tion within the smaller grains than the larger grains. This isconsistent with a previous research (Ref 17) showing that withgrain refinement in a ferrite-martensite DP steel, despite theincrement of material strength, less severe stress/strain parti-tioning between ferrite and martensite would occur. This couldlead to enhanced martensite plasticity and better interfacecohesion that would increase the void nucleation resistance inthese grains.

On the other hand, in large ferrite grains incorporating alower amount of ferrite-martensite interfaces per unit volume,the mentioned back stresses would be lower (Ref 21); So, lowerstresses needed to generate dislocations resulted in higherdislocation pile-ups within the large ferrite grain and a gradientof dislocations; therefore, the resulting misorientation fromcenter toward the boundaries reached its maximum value at theboundaries (Ref 16). Therefore, the larger the grains size, thehigher the stress concentration at martensite particles imposinghigher shear stress at ferrite-martensite interface. As a result, ahigher possibility of void nucleation would be expected in theseregions, i.e., ferrite-martensite interfaces surrounded by largeferrite grains. Similarly, it was shown that (Ref 22) the largerthe size of martensite particles, the higher the probability oftheir cracking. On the contrary, in a ferrite matrix composed ofdistributed fine martensite particles, martensite cracking was

Fig. 9 Shear bands within large ferrite grains

Fig. 10 GOS map of deformed microstructure

Journal of Materials Engineering and Performance Volume 24(1) January 2015—57

rare and the main void nucleation mechanism was thedecohesion of ferrite- martensite interfaces.

To summarize, in a ferrite-martensite DP steel, the largerthe ferrite grain size, or the lower the restricting effect ofmartensite particles surrounding a ferrite grain, the easier isthe shear deformation and the creation of dislocation pile-upswithin the ferrite grain. Consequently, strain distributionwould be less uniform within the grain; this makes voidformation easier at the interface of the larger ferrite grainscompared to the smaller grains. It can be concluded thatsimultaneously decreasing both ferrite and martensite grainsize leads to better strain distribution, lower void nucleationprobability, and finally, higher void nucleation resistancewithin the DP microstructure.

4. Conclusions

In the present research, a high-strength structural steel,DP780, was mechanically tested under room temperatureuniaxial tensile test condition. Detailed microstructural SEMand EBSD analysis revealed the mechanism of void nucleation.This can be summarized as follows:

1. The constraint effect of martensite particles on ferritegrains during deformation imposed rather pronouncedback stresses within small ferrite grains, promoting moreuniform strain distribution, lower GOS as well as higherGAM values within the smaller ferrite grains comparedto the larger grains.

2. The constraint effect of martensite particles on ferritegrains during deformation resulted in lower shear defor-mation ability as well as less dislocation pile-ups in thesmaller ferrite grains compared to the larger grains.

3. More uniform strain distribution and lower shear defor-mation ability as well as the lower number of dislocationpile-ups at the ferrite-martensite interfaces lowered thepossibility of void nucleation in the smaller grains thanthat in the larger grains.

References

1. H. Jin, W.Y. Ln, J.W. Foulk, A. Mota, G. Johnson, and J. Korellis, AnExamination of Anisotropic Void Evolution in Aluminum Alloy 7075,Exp. Mech., 2013, 53, p 1583–1596

2. J. Pospiech, Ductile Fracture of Carbon Steels: A Review, J. Mater.Eng. Perform., 1995, 4, p 82–89

3. J.K. Cuddy and M.N. Bassim, Ductile Fracture Mechanisms in AISI,4340 Steel, Mater. Sci. Eng. A., 1990, 125, p 43–48

4. E. Ahmad, T. Manzoor, M.M.A. Ziai, and N. Hussain, Effect ofMartensite Morphology on Tensile Deformation of Dual-Phase Steel, J.Mater. Eng. Perform., 2012, 21, p 382–387

5. G. Avramovic-Cingara, Y. Ososkova, M.K. Jain, and D.S. Wilkinson,Effect of Martensite Distribution on Damage Behavior in DP600 DualPhase Steels, Mater. Sci. Eng. A., 2009, 516, p 7–16

6. G. Avramovic-Cingara, Y. Ososkova, M.K. Jain, and D.S. Wilkinson,Void Nucleation and Growth in Dual-Phase Steel 600 During UniaxialTensile Testing, Metall. Mater. Trans. A., 2009, 40, p 3117–3127

7. J. Kadkhodapour, A. Butz, and S. Ziaei Rad, Mechanisms of VoidFormation During Tensile Testing in a Commercial, Dual-Phase Steel,Acta. Mater., 2011, 59, p 2575–2588

8. M. Calcagnotto, D. Ponge, and D. Raabe, Effect of Grain Refinementto 1 lm on Strength and Toughness of Dual-Phase Steels, Mater. Sci.Eng. A., 2010, 527, p 7832–7840

9. X. Pan, X. Wu, K. Mo, X. Chen, J. Almer, J. Ilavsky, D.R. Haeffner,and J.F. Stubbins, Lattice Strain and Damage Evolution of 9-12% CrFerritic/Martensitic Steel During In Situ Tensile Test by X-RayDiffraction and Small Angle Scattering, Nucl. Mater., 2010, 407, p10–15

10. K. Kocatepe, M. Cerah, and M. Erdogan, The Tensile FractureBehavior of Intercritically Annealed and Quenched Tempered FerriticDuctile Iron with Dual Matrix Structure, Mater. Des., 2007, 28, p 172–181

11. M. Sarwar, E. Ahmad, K.A. Qureshi, and T. Manzoor, Influence ofEpitaxial Ferrite on Tensile Properties of Dual Phase Steel,Mater. Des.,2007, 28, p 335–340

12. Annual book of ASTM Standards, A370, 200113. K. Fujiyama, K. Mori, D. Kaneko, H. Kimachi, T. Saito, R. Ishii, and T.

Hino, Creep Damage Assessment of 10Cr-1Mo-1W-VNbN SteelForging Through EBSD Observation, Int. J. Pres. Ves. Pip., 2009,86, p 570–577

14. K. Fujiyama, K. Mori, D. Kaneko, H. Kimachi, T. Saito, R. Ishii, and T.Hino, Creep-Damage Assessment of High Chromium Heat ResistantSteels and Weldments, Mater. Sci. Eng. A., 2009, 510–511, p 195–201

15. P.A. Karnezis, G. Durrant, and B. Cantor, Characterization ofReinforcement Distribution in Cast Al-Alloy/SiCp Composites, Mater.Charact., 1988, 40, p 97–109

16. N. Saeidi, F. Ashrafizadeh, B. Niroumand, and F. Barlat, Evaluation ofFracture Micromechanisms in a Fine Grained Dual Phase Steel DuringUniaxial TensileDeformation, J. Iron Steel Res. Int., 2014, 84, p 1386–1392

17. M. Calcagnotto, D. Ponge, and D. Raabe, Deformation and FractureMechanisms in Fine- and Ultrafine-Grained Ferrite/Martensite Dual-Phase Steels and the Effect of Aging, Acta. Mater., 2011, 59, p 658–670

18. J. Kadkhodapour, S. Schmauder, D. Raabe, S. Ziaei Rad, U. Weber,and M. Calcagnotto, Experimental and Numerical Study on Geomet-rically Necessary Dislocations and Non-homogeneous MechanicalProperties of the Ferrite Phase in Dual Phase Steels, Acta. Mater., 2011,59, p 4387–4394

19. S. Lefebvre, B. Devincre, and T. Hoc, Yield Stress Strengthening inUltrafine-Grained Metals: A Two-Dimensional Simulation of Disloca-tion Dynamics, J. Mech. Phys. Solids., 2007, 55, p 788–802

20. L. Zhonghua and G. Haicheng, Bauschinger Effect and Residual PhaseStresses in Two Ductile-Phase Steels: Part I. The Influence of PhaseStresses on the Bauschinger Effect, Metall. Trans. A., 1990, 21, p 717–724

21. A.L. Helbert, X. Feaugas, and M. Clavel, Effects of MicrostructuralParameters and Back Stress on Damage Mechanisms in a/b TitaniumAlloys, Acta. Mater., 1998, 46, p 939–951

22. M. Erdogan and S. Tekeli, The Effect of Martensite Particle Size onTensile Fracture of Surface-Carburised AISI, 8620 Steel with DualPhase Core Microstructure, Mater. Des., 2002, 23, p 597–604

58—Volume 24(1) January 2015 Journal of Materials Engineering and Performance

EBSD Study of Damage Mechanisms in a High-Strength Ferrite-Martensite Dual-Phase SteelAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionsReferences